The most common piece of feedback from our beta users was simple: "The content is good, but it doesn't quite sound like me." We took that seriously.
This update focuses entirely on output quality and voice fidelity — making SermonAI-generated content feel less like something a machine wrote and more like something you wrote on a good day.
Transcription improvements
We've significantly improved how SermonAI handles the kinds of words and phrases that appear in sermons:
- Scripture references — book names, chapter and verse numbers, and common abbreviations are now recognized more reliably
- Proper nouns — names of people, places, and theological terms transcribe more accurately
- Preaching cadence — repeated phrases, emphasis patterns, and sermon-style rhythm are preserved better in the transcript
More accurate transcription means better raw material for every content type downstream.
Voice and tone fidelity
The AI now spends more time analyzing the patterns in your specific sermon before generating outputs. The result:
- Blog posts read more like you wrote them, not like a generic AI article
- Sermon summaries preserve the emotional arc and key illustrations from your message
- Daily devotions carry your pastoral tone across all seven days
Regenerate without losing other outputs
You can now regenerate any single content type without affecting the others. If your blog post needs a fresh take, regenerate just that — your discussion questions and verse list stay untouched.
Thank you to everyone who sent feedback during the first weeks of beta. Keep it coming.